Warning! Long post.
The rise of market power by De Loecker and Eeckhout is creating a kerfuffle in econland. Coming on the heels of Autor's finding of growing industry concentration as well as Barkai's finding of rising "economic" profits, a narrative of growing monopoly power is starting to gather steam. Popular blogger and Bloomberg columnist, Noah Smith, has called for trust-busting to restore economic growth and productivity. Tyler Cowen and like-minded economists have quibbled with the findings. Rising market power has potential political ramifications because it is now starting to gain traction on both sides of the aisle (I saw a Fox News commentator the other day decrying the market power of Google and Facebook and calling for antitrust action). Be that as it may, I want to discuss some theoretical issues I have with the paper, shed light on some non-issues, and give my own take on why mark-ups have risen. Briefly,
1) Calculation of mark-ups over marginal cost has serious theoretical problems. Nonetheless, average mark-up, based on NIPA, does show a rising trend beginning in the 2000s--but not in the 1980s.
2) Rising fixed costs are not the culprit (Tyler's main point). It is true that fixed costs have risen, but the trend of rising fixed costs long predates the rise in mark-up. More important, the biggest rise in mark-up has occurred during the past 15 years when fixed costs' share has been virtually stagnant.
3) I am not convinced that rising industry concentration is the driving force.
4) My take is that large US firms with dominant brands/market positions have always commanded monopolistic rents. What has changed is that the rents, which used to be more broadly shared by stakeholders, are now predominantly flowing to shareholders and top management. The change has been driven by the interplay of several developments--shareholder revolution, globalization, rising equity valuations, diminished growth expectations.
Theoretical Issues
Dietrich Vollrath has an excellent blog post where he brings up some of the theoretical issues with the paper. Among these, I find the assumption of unchanging production function through time the most questionable.
Aside from the issues highlighted by DV, I have a few theoretical quibbles.
First, there is long literature on whether firms optimize at all, starting with the empirical work of Hall and Hitch (1939). Alternate theories of firms behavior--managerial and behavioral--have an equally hoary history staring with at least Berle and Means (1932).
Second, I do think that the shareholder revolution triggered by Jensen and Meckling (1976) has changed firm behavior meaningfully. Remember Gordon Gekko's speech about 37 vice presidents with no stake! If so, firms' objective functions have probably changed through time and the assumption of cost minimization through history is questionable.
Third, the shift away from the conglomerates of the 190s and 1970s toward more focused firms surely makes the identification of fixed and variable costs complicated.
Fourth, the categories of variable and fixed costs are not iron-clad. Obviously, over long enough time frames, all costs are variable. Leaving that aside, even production-line workers have firm-specific capital. So, the costs related to training them or firing them are not entirely variable. More important, the question is how businesses view employee costs. There is circumstantial evidence that businesses have gone from treating most employees as relatively fixed costs to relatively variable costs. Note that since 1980, the proportion of job losers as a share of those unemployed has risen, suggesting that firms increasingly view employees as a variable cost.
Mark-Ups Based on NIPA Data
D&E use Compustat data, which pertains to only listed firms. Let us take a look at NIPA data, which covers all incorporated businesses operating in the United States. We have to make do with mark-up over average costs rather than marginal costs. However, NIPA data are comprehensive and capitalize R&D and software development costs--both of which have become increasingly important in recent decades.
First, gross value value added as a % of employee costs (a proxy for average mark-up) has indeed increased, However, that rise can only be traced to 2001.
Second, taking depreciation (which in the NIPA includes amortization of software, R&D, etc) as a proxy for fixed costs, fixed costs as a share of gross value added has risen. However, note that the trend of rising fixed costs long predates 1980, which the demarcation line for rising mark-ups. So, while fixed costs have risen as Tyler Cowen argues, they do not appear to be the reason for the rise in mark-up. Interestingly, note that depreciation as a share has really flattened since 2001--the period in which the NIPA data show the biggest rise in mark-up!
Rising Industry Concentration?
Although Autor et al have shown rising industry concentration since 1980, Grullon et al show industry concentration (HHI) rising only since 1997 (chart below). Moreover, industry concentration is only back to the levels of early 1980s. At any rate, the Grullon story is more consistent with the mark-up picture portrayed by the NIPA for the past 15-16 years.
I have to do more research but my impression is that U.S. industry has not hugely changed through time.
What's Behind Rising Mark-Ups?
Large U.S. firms have always commanded monopolistic rents--think Dupont, Bethlehem Steel, IBM, GM/Ford/Chrysler in their respective heydays. However, several developments have worked to dramatically change how those rents are shared.
Before the 1980s shareholder revolution, monopolistic rents of dominant firms were more broadly shared--not just with rank and file employees but with the local community. For instance, Bethlehem Steel at its peak lavishly funded the local community as did many other similar firms. AT&T's and IBM ran massive research operations, funding basic research that would be inconceivable in today.
When dominant firms' rents are more broadly shared, it also creates less dispersion of profits among firms. To see why this happens, we need to recall circular flow. Employees of one firm spend their incomes on products of other firms. If employee compensation share in dominant firms declines, then the revenue accruing to the rest of the corporate sector weakens, reducing their profits. In other words, greater extraction of dominant company rents by shareholders and top managements ends up creating the winner-take-all distribution of profits.
Globalization probably has played some role in the increase in mark-up. First, the shift of low-end manufacturing means that what is left is high-end manufacturing with greater monopolistic power. So, mark-ups should increase simply from that shift. Unsurprisingly, the NIPA mark-up show that the rise in mark-up coincided with China's entry into WTO and is also consistent with Autor's findings about impact of Chinese competition. Second, if firms already have some pricing power, reducing costs through outsourcing should result in higher mark-ups.
Of course, there is a fallacy of composition in the last assertion. Cutting labor costs means less revenue flowing to the corporate sector. However, if workers make up for lost income by taking on more debt to maintain spending--which certainly has happened in the US--and the government supports household consumption via increased transfers--which too has occurred--then the fallacy of composition is resolved.
Lastly, high equity valuations combined with diminished growth expectations have exerted enormous pressures on business executives to maintain high margins to justify valuations. The 2001 recession is often termed as a mild recession or a non-recession by many economists. But in my opinion many of our problems can be traced to that recession. Job market churn remains depressed as companies have become increasingly reluctant to expend significant resources on employee training. This reluctance also manifests in the yawning gap between job openings and hirings. All of these can be traced to the pressure to maintain margins in an anemic growth environment.
I will stop here. It is already too long!
The rise of market power by De Loecker and Eeckhout is creating a kerfuffle in econland. Coming on the heels of Autor's finding of growing industry concentration as well as Barkai's finding of rising "economic" profits, a narrative of growing monopoly power is starting to gather steam. Popular blogger and Bloomberg columnist, Noah Smith, has called for trust-busting to restore economic growth and productivity. Tyler Cowen and like-minded economists have quibbled with the findings. Rising market power has potential political ramifications because it is now starting to gain traction on both sides of the aisle (I saw a Fox News commentator the other day decrying the market power of Google and Facebook and calling for antitrust action). Be that as it may, I want to discuss some theoretical issues I have with the paper, shed light on some non-issues, and give my own take on why mark-ups have risen. Briefly,
1) Calculation of mark-ups over marginal cost has serious theoretical problems. Nonetheless, average mark-up, based on NIPA, does show a rising trend beginning in the 2000s--but not in the 1980s.
2) Rising fixed costs are not the culprit (Tyler's main point). It is true that fixed costs have risen, but the trend of rising fixed costs long predates the rise in mark-up. More important, the biggest rise in mark-up has occurred during the past 15 years when fixed costs' share has been virtually stagnant.
3) I am not convinced that rising industry concentration is the driving force.
4) My take is that large US firms with dominant brands/market positions have always commanded monopolistic rents. What has changed is that the rents, which used to be more broadly shared by stakeholders, are now predominantly flowing to shareholders and top management. The change has been driven by the interplay of several developments--shareholder revolution, globalization, rising equity valuations, diminished growth expectations.
Theoretical Issues
Dietrich Vollrath has an excellent blog post where he brings up some of the theoretical issues with the paper. Among these, I find the assumption of unchanging production function through time the most questionable.
Aside from the issues highlighted by DV, I have a few theoretical quibbles.
First, there is long literature on whether firms optimize at all, starting with the empirical work of Hall and Hitch (1939). Alternate theories of firms behavior--managerial and behavioral--have an equally hoary history staring with at least Berle and Means (1932).
Second, I do think that the shareholder revolution triggered by Jensen and Meckling (1976) has changed firm behavior meaningfully. Remember Gordon Gekko's speech about 37 vice presidents with no stake! If so, firms' objective functions have probably changed through time and the assumption of cost minimization through history is questionable.
Third, the shift away from the conglomerates of the 190s and 1970s toward more focused firms surely makes the identification of fixed and variable costs complicated.
Fourth, the categories of variable and fixed costs are not iron-clad. Obviously, over long enough time frames, all costs are variable. Leaving that aside, even production-line workers have firm-specific capital. So, the costs related to training them or firing them are not entirely variable. More important, the question is how businesses view employee costs. There is circumstantial evidence that businesses have gone from treating most employees as relatively fixed costs to relatively variable costs. Note that since 1980, the proportion of job losers as a share of those unemployed has risen, suggesting that firms increasingly view employees as a variable cost.
Mark-Ups Based on NIPA Data
D&E use Compustat data, which pertains to only listed firms. Let us take a look at NIPA data, which covers all incorporated businesses operating in the United States. We have to make do with mark-up over average costs rather than marginal costs. However, NIPA data are comprehensive and capitalize R&D and software development costs--both of which have become increasingly important in recent decades.
First, gross value value added as a % of employee costs (a proxy for average mark-up) has indeed increased, However, that rise can only be traced to 2001.
Second, taking depreciation (which in the NIPA includes amortization of software, R&D, etc) as a proxy for fixed costs, fixed costs as a share of gross value added has risen. However, note that the trend of rising fixed costs long predates 1980, which the demarcation line for rising mark-ups. So, while fixed costs have risen as Tyler Cowen argues, they do not appear to be the reason for the rise in mark-up. Interestingly, note that depreciation as a share has really flattened since 2001--the period in which the NIPA data show the biggest rise in mark-up!
Rising Industry Concentration?
Although Autor et al have shown rising industry concentration since 1980, Grullon et al show industry concentration (HHI) rising only since 1997 (chart below). Moreover, industry concentration is only back to the levels of early 1980s. At any rate, the Grullon story is more consistent with the mark-up picture portrayed by the NIPA for the past 15-16 years.
I have to do more research but my impression is that U.S. industry has not hugely changed through time.
What's Behind Rising Mark-Ups?
Large U.S. firms have always commanded monopolistic rents--think Dupont, Bethlehem Steel, IBM, GM/Ford/Chrysler in their respective heydays. However, several developments have worked to dramatically change how those rents are shared.
Before the 1980s shareholder revolution, monopolistic rents of dominant firms were more broadly shared--not just with rank and file employees but with the local community. For instance, Bethlehem Steel at its peak lavishly funded the local community as did many other similar firms. AT&T's and IBM ran massive research operations, funding basic research that would be inconceivable in today.
When dominant firms' rents are more broadly shared, it also creates less dispersion of profits among firms. To see why this happens, we need to recall circular flow. Employees of one firm spend their incomes on products of other firms. If employee compensation share in dominant firms declines, then the revenue accruing to the rest of the corporate sector weakens, reducing their profits. In other words, greater extraction of dominant company rents by shareholders and top managements ends up creating the winner-take-all distribution of profits.
Globalization probably has played some role in the increase in mark-up. First, the shift of low-end manufacturing means that what is left is high-end manufacturing with greater monopolistic power. So, mark-ups should increase simply from that shift. Unsurprisingly, the NIPA mark-up show that the rise in mark-up coincided with China's entry into WTO and is also consistent with Autor's findings about impact of Chinese competition. Second, if firms already have some pricing power, reducing costs through outsourcing should result in higher mark-ups.
Of course, there is a fallacy of composition in the last assertion. Cutting labor costs means less revenue flowing to the corporate sector. However, if workers make up for lost income by taking on more debt to maintain spending--which certainly has happened in the US--and the government supports household consumption via increased transfers--which too has occurred--then the fallacy of composition is resolved.
Lastly, high equity valuations combined with diminished growth expectations have exerted enormous pressures on business executives to maintain high margins to justify valuations. The 2001 recession is often termed as a mild recession or a non-recession by many economists. But in my opinion many of our problems can be traced to that recession. Job market churn remains depressed as companies have become increasingly reluctant to expend significant resources on employee training. This reluctance also manifests in the yawning gap between job openings and hirings. All of these can be traced to the pressure to maintain margins in an anemic growth environment.
I will stop here. It is already too long!
